error-recovery

Handle transient API failures in distributed agent workflows by applying automated retry with backoff, fallback alternatives, and graceful degradation strategies to maintain operation during partial service unavailability.

Fault tolerance for squad agents is achieved through adaptive recovery patterns that automatically switch to fallback tools, diagnose errors, and escalate with context when primary services time out or become unavailable.

Graceful degradation for distributed systems lets agents continue delivering value by skipping non-critical steps during failures, documenting what was skipped, and applying retry with backoff for transient errors.

Automated error-recovery patterns support both retry logic with increasing delays and escalation with context, summarizing all attempts and errors for human review when adaptive recovery strategies are exhausted.

Use diagnose-and-fix recovery instead of simple retry with backoff when errors require targeted analysis and proposed fixes, allowing agents to analyze errors, apply corrections, and verify results before escalating.

Error-recovery operates without external dependencies, providing standard resilience patterns including retry with backoff, fallback alternatives, and clear logging to minimize downtime across distributed agent tasks.